• Wed. Nov 22nd, 2023

    Critical Thought

    Critical thoughts on quantum technologies

    The Future of Quantum Machine Learning in Finance: A Promising Path Forward

    ByByron Bekker

    Nov 13, 2023
    The Future of Quantum Machine Learning in Finance: A Promising Path Forward

    In a recent interview with Javier Mancilla, a QML researcher at Stafford Computing, new insights into the potential of quantum computing in the finance sector have emerged. Mancilla delves into the challenges and opportunities presented by quantum machine learning, offering valuable perspectives on its practical application in the industry.

    When asked about the timing of when quantum machine learning will become a practical production tool, Mancilla highlights the significant investments made in quantum computing since 2018. He emphasizes that while NISQ machines are not yet ready for market-level operations due to limitations like noise and errors, there is still potential to harness the power of quantum software. Quantum software enables the utilization of simulators and emulators, providing valuable insights that can be applied in finance and other fields.

    Mancilla expresses the need for quick wins in the quantum computing industry and suggests exploring the use of simulators in the meantime. While gate-based quantum computing may not currently offer practical solutions for hardware applications, simulators can offer a glimpse into the potential of quantum approaches. By leveraging quantum machine learning techniques, such as quantum support vector classifiers and quantum neural networks, businesses can gain a “quantum business advantage” by surpassing their existing stack of solutions.

    In response to questions about the sufficiency of current simulators, Mancilla explains that while they may be limited to around 30-50 qubits, they still hold significant value for certain applications. He notes that optimization problems, which require more qubits, can be addressed by platforms like D-Wave. In fact, D-Wave’s feature selection optimization has shown promising results. However, for classification models and tasks like fraud detection and churn prediction, Mancilla highlights the effectiveness of variational classifiers even with a smaller number of qubits. He presents a practical example where dimensionality reduction and classification were achieved using just two qubits, demonstrating the potential impact of even a limited qubit range.

    Regarding the concept of “barren plateaus” in quantum machine learning models, Mancilla acknowledges that reaching a plateau at 20-22 qubits can be discouraging. However, he emphasizes that this does not diminish the value of quantum-inspired algorithms. These algorithms can be run on classical computers, leveraging techniques like tensor networks to address real-life use cases effectively.

    As the finance sector explores the application of quantum machine learning, Javier Mancilla’s insights provide valuable guidance for researchers and industry professionals. While challenges remain, the promise of quantum computing in finance is a beacon of hope, and through continued research and development, the future of quantum machine learning looks increasingly bright.

    Frequently Asked Questions (FAQ)

    Q: What is quantum machine learning?

    A: Quantum machine learning refers to the utilization of quantum computing and quantum algorithms to enhance traditional machine learning methods. It explores the potential of quantum systems to process and analyze large datasets more efficiently and solve complex optimization problems.

    Q: What are NISQ machines?

    A: NISQ (Noisy Intermediate-Scale Quantum) machines are quantum computers that are currently available for research and development purposes. They are characterized by the presence of noise and errors in their quantum operations, which limits their reliability and scalability for practical applications.

    Q: What are simulators and emulators in quantum computing?

    A: Simulators and emulators are software tools that enable the simulation and emulation of quantum computers on classical hardware. They allow researchers and developers to explore quantum algorithms, test their performance, and gain insights without the need for physical quantum hardware.